X. Ding1, E.L. Abner3,5,6, F.A. Schmitt3,4, J. Crowley7, P. Goodman8, R.J. Kryscio2,3,5
1. Western Kentucky University, Department of Public Health, Bowling Green, Kentucky, USA; 2. University of Kentucky, Department of Statistics, Lexington, Kentucky, USA; 3. University of Kentucky, Sanders-Brown Center on Aging, Lexington, Kentucky, USA; 4. University of Kentucky, Department of Neurology, Lexington, Kentucky, USA; 5. University of Kentucky, Department of Biostatistics, Lexington, Kentucky, USA; 6. University of Kentucky, Department of Epidemiology, Lexington, Kentucky, USA; 7. SWOG Cancer Research and Biostatistics, Seattle, WA, USA; 8. SWOG Statistical Center, Fred Hutchinson Cancer Research Center, Seattle, WA, USA.
Corresponding Author: Xiuhua Ding, M.D., Ph.D., Department of Public Health, Western Kentucky University, 1906 College Heights Blvd, Bowling Green, KY 42101, USA, Email: email@example.com, phone: 270-745-3618, Fax: 270-745-6950
J Prev Alz Dis 2020;
Published online September 18, 2020, http://dx.doi.org/10.14283/jpad.2020.50
Background: The Medical Outcomes Study Questionnaire Short Form 36 health survey (SF-36) measures health-related quality of life (HRQoL) from the individual’s point of view and is an indicator of overall health status.
Objective: To examine whether HRQoL shows differential changes over time prior to dementia onset and investigate whether HRQoL predicts incidence of dementia.
Design: Prevention of Alzheimer’s Disease (AD) by Vitamin E and Selenium (PREADViSE) trial, which recruited 7,547 non-demented men between 2002 and 2009. A subset of 2,746 PREADViSE participants who completed up to five SF-36 assessments at annual visits was included in the current analysis
Setting: Secondary data analysis of PREADViSE data.
Participants: A subset of 2,746 PREADViSE participants who completed up to five SF-36 assessments at annual visits was included in the current analysis.
Measurements: Two summary T scores were generated for analysis: physical component score (PCS) and mental component score (MCS), each with a mean of 50 (standard deviation of 10); higher scores are better. Linear mixed models (LMM) were applied to determine if mean component scores varied over time or by eventual dementia status. Cox proportional hazards regression was used to determine if the baseline component scores were associated with dementia incidence, adjusting for baseline age, race, APOE-4 carrier status, sleep apnea, and self-reported memory complaint at baseline.
Results: The mean baseline MCS score for participants who later developed dementia (mean± SD: 53.9±9.5) was significantly lower than for those participants who did not develop dementia during the study (mean±SD: 56.4±6.5; p = 0.005). Mean PCS scores at baseline (dementia: 49.3±7.9 vs. non-dementia: 49.8±7.8) were not significantly different (p = 0.5) but LMM analysis showed a significant time effect. For MCS, the indicator for eventual dementia diagnosis was significantly associated with poorer scores after adjusting for baseline age, race, and memory complaint. Adjusted for other baseline risk factors, the Cox model showed that a 10-unit increase in MCS was associated with a 44% decrease in the hazard of a future dementia diagnosis (95% CI: 32%-55%).
Conclusion: The SF-36 MCS summary score may serve as a predictor for future dementia and could be prognostic in longitudinal dementia research.
Key words: Health-related Quality of Life, dementia, outcome measures.
Dementia is not a specific disease but a syndrome defined by its symptoms (1). Alzheimer’s disease (AD) is the most common form of dementia, and worldwide around 50 million people live with AD and other dementias; over 5 million with dementia live in the U.S. (2). These numbers are expected to grow as the population ages, but the estimated annual costs of dementia exceeded $600 billion (U.S. dollars) already by 2010 (3). With increasing prevalence, these costs are expected to increase by 85% by 2030 (3). Therefore, AD and related dementias are important public health issues. With no cure currently available and with anticipated increases in both prevalence and costs, early diagnosis and intervention are keys to maintaining health and independence for as long as possible. .
Health-related quality of life (HRQoL) describes one’s physical, mental, emotional, and social functioning (4-6). The Medical Outcomes Study 36-item Short Form (SF-36) survey is one of most commonly used instruments to measure HRQoL (7). Although many previous studies used SF-36 to examine HRQoL for participants with chronic diseases (8-10), SF-36 HRQoL scores are also associated with mortality risk and incidence of disease: Nilsson et al. used the SF-36 to predict coronary heart disease incidence in a middle-age Swedish population (11). Drageset et al. found that HRQoL predicted cancer mortality for nursing home residents without cognitive impairment (12).
The SF-36 is a general health measure with eight domains: Physical Functioning (PF), Role Limitations Due to Physical Problems (RP), Bodily Pain (BP), General Health Perceptions (GH), Vitality (VT), Social Functioning (SF), Role Limitations Due to Emotional Problems (RE), and General Mental Health (MH). Heterogeneous methods have been used to analyze and interpret SF-36 scores. Several studies have examined SF-36 measurements by comparing a single SF-36 domain score between different groups (13, 14), while others have created a total score. However, Lins et al. suggest that using a single SF-36 total score, which necessarily ignores potential differences in physical and mental HRQoL, may lead to poor validity (15).
The SF-36’s eight domains can be clustered into two component summary scores: a physical component summary score (PCS) and a mental component summary score (MCS). The SF-36 user’s manual introduced two ways to develop these summary scores: psychometric-based summary measures and standardized scoring (norm-based summary measures). For the latter method, the manual (16) provides the necessary documentation to score and interpret the PCS and MCS. Studies report that SF-36 summary scores account for 80-85% of variance in the eight SF-36 domain scales, and thus their use in place of the eight domains can reduce the number of statistical comparisons needed (17). The PCS and MCS summary scores are expected to perform reliably in both cross-sectional and longitudinal settings (17, 18) and have been used in several studies (11, 12, 14, 19).
The aims of this study were to investigate whether SF-36 PCS and MCS scores change over a five-year interval and to determine whether the baseline SF-36 PCS and MCS scores are associated with risk of dementia in a large sample of older adult men who were enrolled in a randomized, double-blind clinical trial to prevent dementia.
Study population and data sources
This secondary data analysis is based on a subset of participants from the Prevention of Alzheimer’s Disease by Vitamin E and Selenium (PREADViSE) trial (20, 21). PREADViSE was an ancillary study to the Selenium and Vitamin E Cancer Prevention Trial (SELECT) (NCT00006392) and recruited 7,547 men without dementia age 60 and older from 128 participating SELECT (a large prostate cancer prevention RCT) sites to assess the effectiveness of antioxidant supplements in preventing incident Alzheimer’s Disease (AD). SELECT randomized participants to treatments (22): 400 International Units (IU) Vitamin E, 200 micrograms(mg) of Selenium, 400 IU Vitamin E plus 200 mg of Selenium, or Placebo. Since the use of antioxidant supplements did not affect the risk of dementia, the effects of the antioxidant supplements will not be considered further in this report (23).
PREADViSE investigators were blind to SELECT treatment assignment during follow-up. The eligibility criteria for participating in the PREADViSE trial included active SELECT enrollment at a participating site and absence of dementia and other active conditions that affect cognition, such as major psychiatric disorders, including depression. In 2008, the SELECT Data Safety Monitoring Committee discontinued study supplements after a futility analysis for the prostate cancer outcome. PREADViSE then continued as an observational study until 2015 to ascertain incident dementia cases (21). The University of Kentucky Institutional Review Board (IRB) and the IRBs at each SELECT study site approved PREADViSE research activities. All participants provided written informed consent.
Participants were included in this retrospective analysis if they had completed the SF-36 at least once at their annual in-person visits during the randomized clinical trial (RCT) phase of the study (between 2002 and 2009). In total, 2,748 participants out of 7,547 completed up to five SF-36 assessments at annual visits and were included in the current analysis.
The Memory Impairment Screen (MIS) was the primary dementia screening instrument in both the RCT and the observational period of PREADViSE (24, 25). If participants failed the MIS (scored ≤5/8 on the immediate or delayed recall portion of the MIS), a second-tier screen was administered. An expanded Consortium to Establish a Registry in Alzheimer’s Disease battery (CERAD-e) was used during the RCT period (26), and the modified Telephone Interview for Cognitive Status (TICS-m) was used during the observational period (27, 28). The CERAD-e and the TICS-m assessed participants’ global cognitive function. Failure on the second screen (CERAD-e T score ≤ 35, TICS-m total score ≤ 35) led to a recommendation for a clinic visit with their local physician. Three to five expert clinicians, including two neurologists and at least one neuropsychologist, reviewed records from the clinic visit for a consensus diagnosis. In cases in which the neurologists disagreed in their diagnoses, the study primary investigator made the final determination (23). Annual screenings were completed in May 2014, and a small number of participants were followed for medical records through August 2015.
Incident dementia cases were identified using two methods. First, as described above, a medical records-based consensus diagnosis was conducted with date of diagnosis assigned as the date of the failed screen. Second, because many participants were reluctant to obtain medical examination for their memory, additional measures including the AD8 Dementia Screening Interview (29) were employed. In addition to the AD8, dementia determination was based on self-reported medical history; self-reported diagnosis of dementia; use of memory-enhancing prescription drugs; and cognitive scores on the MIS, CERAD-e T score, New York University Paragraph Delayed Recall, and TICS-m. The diagnostic criteria for the second method were AD8 total of 1 or greater (at any time during follow-up) to indicate functional impairment plus one or more of the following: self-reported diagnosis of dementia, use of a memory-enhancing prescription drug (donepezil, rivastigmine, galantamine, memantine), or a cognitive score below the cutoff for intact cognition on any test (e.g., 1.5 standard deviations below expected performance based on age and education normative data). Date of diagnosis was assigned to the earliest event (29).
Health Related Quality of Life
Scoring was performed according to the SF-36 Health Survey Manual and Interpretation Guide (16). First, a raw score for each of the eight domains was calculated by summing the item responses within each domain. Raw summary scores were then transformed to scale scores that ranged from 0 to 100 (16). Each SF-36 domain scale score was standardized to the general U.S. older adult male population ages 65 years and older by computing a z score for each domain (16). Means and standard deviations used to generate the Z scores are given in Supplemental Table 1.
*All PREADViSE participants are male; †SD=standard deviation.
The PCS and MCS were then generated using a weighted sum of the domain Z scores. Aggregate physical and mental component scores were then calculated using the formulas below (16):
PCS = (PF_Z * 0.42402) + (RP_Z * 0.35119) + (BP_Z * 0.31754) + (GH_Z * 0.24954) + (VT_Z * 0.02877) + (SF_Z * -0.00753) + (RE_Z * -0.19206) + (MH_Z * -0.22069)
MCS = (PF_Z * -0.22999) + (RP_Z * -0.12329) +(BP_Z * -0.09731) + (GH_Z * -0.01571) + (VT_Z * 0.23534) + (SF_Z * 0.26876) + (RE_Z * 0.43407) + (MH_Z * 0.48581)
Finally, the PCS and MCS scores were transformed to T scores by multiplying the PCS and MCS sum scores by 10 and adding 50. Then these two component T scores (PCS and MCS) were compared to national normative data based on the manual (16) and were used in statistical modelling.
Data were also collected on age at baseline, race, ethnicity, years of education, APOE ε4 allele carrier status, and self-reports on the presence of comorbidities including diabetes mellitus, hypertension, sleep apnea, memory complaint (measured as “Have you noticed any changes in your memory?”), and family history of dementia in a first degree relative. These covariates were selected based on results from the descriptive analysis and findings from previous studies (20).
Chi-square and t-test statistics were used compare categorical and continuous variables between participants who did not and did not develop dementia during follow-up. And general characteristics between PREADViSE participants who completed SF-36 and PREADViSE participants who did not complete SF-36. Linear mixed models (LMM) were constructed for SF-36 PCS and MCS summary scores, with time-dependent between-subjects factor cognitive status at the year of assessment (dementia versus non-dementia) and within-subjects factor year of assessment; a cognitive status*year interaction term was also included. Covariates were age at baseline, years of education, Black race (Yes vs. No), Hispanic ethnicity (Yes vs. No), APOE (presence vs absence of at least one ε4 allele), self-reported baseline indicators for diabetes mellitus, hypertension, memory problem, sleep apnea, and family history of dementia.
To determine if the baseline SF-36 PCS and MCS, which were defined as the PCS and MCS from the first SF-36 measurement in PREADViSE, affected the hazard of dementia, a series of Cox proportional hazards regression models with SF-36 PCS or MCS as the independent variable and survival time to diagnosis of dementia as the dependent variable, were applied to a multivariable survival analysis. The follow-up time was defined as years between date of PREADViSE enrollment and date of dementia diagnosis or, in the absence of dementia, date of last assessment. The multivariable Cox models included age at baseline, Black race (Yes vs. No), APOE (presence vs absence of at least one ε4 allele), sleep apnea at baseline, and self-reported memory complaint and PCS or MCS. Hypertension, diabetes and family history of dementia, years of education were excluded from the model due to insignificance. Since the PCS was not included in the final model due to insignificance, the proportional hazard assumption was tested for MCS only through maximum residual method and it was met.
All data were analyzed using PC-SAS version 9.4, and 0.05 was set as the significance level.
Demographic characteristics of study participants with and without SF-36 data are given in Supplemental Table 2. PREADViSE participants who completed the SF-36 were similar to participants who never completed the SF-36 in terms of the proportion of APOE ε4 carriers, but were slightly younger, less educated, more likely to report Black race or Hispanic ethnicity, and less likely to report family history of dementia (Supplemental Table 2).
Men who developed dementia (n=128) were significantly older at baseline (p <0.001), more likely to report Black race (p=0.01and memory change (p <0.001), but less likely to report Hispanic ethnicity (p<0.001) (Table 1).They were also more likely to carry at least one APOE ε4 allele (p = 0.01) (Table 1). Compared to the population of U.S. males ages 65 and over (16), in this sample mean PCS (mean±SD: 49.8 ± 7.8) was significantly higher than the general population (mean±SD: 42.0 ± 11.4; p<0.001), and mean MCS (mean±SD: 56.3 ± 6.7) in the study sample was also significantly higher than the general population (mean±SD: 52.5 ± 9.8; p<0.001). These differences are expected in a population of healthy men who would be motivated to enroll in a prevention trial. On average, men were followed up 5.9 ± 2.7 years. Men who developed dementia were followed up longer than the men who did not develop dementia (p<0.001) (Table 1).
Men who developed dementia had baseline MCS scores (mean±SD: 53.9 ± 9.5) that were significantly lower than the men who did not develop dementia (mean±SD: 56.4 ± 6.5; p=0.005), while there was not a significant difference in SF-36 PCS baseline scores (mean±SD for dementia: 49.3 ± 7.9 vs. mean±SD for non-dementia: 49.8 ± 7.8; p=0.54). Means for PCS and MCS by cognitive status at each visit are depicted in Figure 1a and Figure 1b, respectively. Mean PCS significantly decreased over time, but no significant difference was observed in the rate of decline between men who developed dementia and men who did not. Lower mean MCS scores were significantly associated with incidence (risk) of dementia, but there was not a significant change over time. These associations remained in our adjusted LMM. The LMM analysis showed no interaction effect between dementia status and year at assessment. There were significant effects of time for PCS and dementia group for MCS, respectively after adjusting for covariates. PCS declined linearly by 0.46 (SE: 0.03) each year (p<0.001). Participants with an eventual dementia diagnosis had lower overall estimated MCS 2.86 (SE: 0.8) than participants who did not develop dementia (p<0.001) (Supplemental Table 3).
Table 2 displays adjusted hazard ratios (HRs) for dementia diagnosis from multivariable Cox models. Baseline SF-36 MCS was significantly associated with risk of dementia in the adjusted model (HR = 0.6 for a 10-unit difference, 95% CI=0.5-0.7), while baseline SF-36 PCS was not significant.
*Variables included in the adjusted models were MCS, Baseline age, Black, APOE ε4 carrier, Memory complaint, Sleep Apnea.
This study investigated whether SF-36 PCS and MCS summary scores were changed over time, and whether they were associated with the incidence/risk of dementia, in a U.S. older adult male population. We found that higher baseline SF-36 MCS was associated with a significantly lower risk of dementia. Baseline PCS was not significantly associated with risk of dementia. Over 5 years of follow-up, PCS decreased significantly but slowly, while MCS did not change significantly.
Although previous studies on the SF-36 as a predictor of incident dementia diagnosis are rare, the result that MCS can predict incidence of dementia is not surprising. There are several possible explanations to describe the associations between SF-36 summary scores and risk of dementia. First, lower MCS score may be a preclinical manifestation for cognitive impairment and/or dementia. The MCS score may capture concerns over early cognitive changes. Second, three scales—Mental Health, Role-Emotional, and Social Functioning—are the most heavily weighted domains that contribute to the MCS composite measure. Lower MCS scores potentially indicate presence of psychological and/or psychosocial risk factors such as social support and networks, exposure to discrimination, satisfaction with the life, which may be highly associated with risk of dementia. Peitsch et al showed that general life satisfaction predicted risk of dementia in older adults (30). Gulpers et al showed through a meta-analysis that anxiety is associated with increased risk for dementia (31). Furthermore, the relationship between lower MCS and increased risk of dementia can be traced back to self-reported health. Several studies reported that self-reported health associates with dementia risk. Montlahuc et al. showed that increased risk of dementia was associated with poor (adjusted HR = 1.7, 95% CI: 1.2-2.4) or fair self-rated health (adjusted HR = 1.5, 95% CI: 1.00-2.2) compared to those with good self-related health (32). Abner et al. found that subjective memory complaints are associated with increased risk of dementia (33). Certainly, cognitively impaired participants may not rate the same set of health conditions in the same way as participants who are not cognitively unimpaired. The lower MCS score may also represent unmeasured or undiagnosed diseases or other unmeasured confounding factors. We would hypothesize that a brain disease may manifest earlier in mental symptoms than physical ones, particularly in the way they are measured by SF-36. Similar to Sabia et al’s study that physical activity was not found as a risk factor of dementia, our study showed that PCS did not predict risk of dementia (34). Also, Shimada et al found that physical frailty was not significantly associated with risk of dementia (35).
This study also has limitations as previously described (20). Only 35.8% PREADiVSE participants completed the SF-36. Participants who completed the SF-36 were similar to participants who never completed the SF-36 on the proportion of APOE 4 carriers, but different on age, education, race, ethnicity, and family history of dementia. So the numbers of cases developed from the participants in this study may be disproportional to the numbers of cases from non-participants, which may lead to biased result. Also, clinical trial participants are often different than the trial’s target population. Here, our sample was highly educated and reported better overall health than the general population of U.S. men age 65 and over, so our results may have limited generalizability to other populations. We plan to replicate this analysis in independent datasets.
Strengths for the current study include a large, well characterized sample with over five years of average follow-up. The use of normative methods to estimate the MCS and PCS SF-36 summary scores, which are easily interpretable, is also a strength. These SF-36 summary scores are also reliable (17). Finally, using the two summary scores reduced the number of statistical comparisons but still allowed for mental and physical HRQoL to differ.
Our study provides evidence that MCS from HRQoL may predict incidence of dementia in older men. Although the exact mechanism remains unclear, this may occur because of unmeasured factors, or underlying neuropsychological factors. In summary, SF-36 MCS may predict future dementia, and thus may have utility either as a modifiable risk factor or early warning sign of impending cognitive decline. Further studies with more diverse populations and longer follow-up are needed.
Funding: Funding Source: PREADViSE (NCT00040378) is supported by NIA R01 AG019421. Additional support for the current study comes from NIA R01 AG038651 and NIA P30 AG028383. SELECT was supported by NCI grants CA37429 and UM1 CA182883. The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of data; in the preparation of the manuscript; or in the review or approval of the manuscript except that NCI was involved in the design of SELECT.
Acknowledgments: We sincerely acknowledge all PREADVISE study participants for their participation and thank all of the PREADVISE support staff and Biostatistics group for assistance with study procedures and data management.
Conflict of interest: Drs. Erin Abner and Richard Kryscio report grants from NIA during the conduct of the study. The other authors have no conflict interests.
Ethical Standards: The University of Kentucky Institutional Review Board (IRB) and the IRBs at each SELECT study site approved PREADViSE research activities. All participants provided written informed consent.
1. Association, A.s., Alzheimer’s Disease & Dementia. 2020.
2. Schaller, S., et al., The main cost drivers in dementia: a systematic review. Int J Geriatr Psychiatry, 2015. 30(2): p. 111-29.
3. Wimo, A., et al., The worldwide economic impact of dementia 2010. Alzheimers Dement, 2013. 9(1): p. 1-11.e3.
4. 2020, H.P., Healthy People 2020.
5. Shaoman Yin, R.N., Lawrence Barker, Paul Z. Siegel, Youlian Liao, Summarizing health-related quality of life (HRQOL): development and testing of a one-factor model. Popul Health Metr., 2016. 14.
6. Prevention, C.f.D.C.a., Measuring healthy days: Population assessment of health-related quality of life. Centers for Disease Control and Prevention , Atlanta, Georgia, 2000.
7. Ware, J.E., Jr. and C.D. Sherbourne, The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care, 1992. 30(6): p. 473-83.
8. McCall, W.V., et al., Effects of continuation electroconvulsive therapy on quality of life in elderly depressed patients: A randomized clinical trial. J Psychiatr Res, 2018. 97: p. 65-69.
9. Moreno, J.A., et al., Caregiving in Dementia and its Impact on Psychological Functioning and Health-Related Quality of Life: Findings from a Colombian Sample. J Cross Cult Gerontol, 2015. 30(4): p. 393-408.
10. Pekmezovic, T., et al., Quality of life in patients with progressive supranuclear palsy: one-year follow-up. J Neurol, 2015. 262(9): p. 2042-8.
11. Nilsson, E., et al., SF-36 predicts 13-year CHD incidence in a middle-aged Swedish general population. Qual Life Res, 2020. 29(4): p. 971-975.
12. Drageset, J., G.E. Eide, and A.H. Ranhoff, Mortality in nursing home residents without cognitive impairment and its relation to self-reported health-related quality of life, sociodemographic factors, illness variables and cancer diagnosis: a 5-year follow-up study. Qual Life Res, 2013. 22(2): p. 317-25.
13. McKee, G., Are there meaningful longitudinal changes in health related quality of life–SF36, in cardiac rehabilitation patients? Eur J Cardiovasc Nurs, 2009. 8(1): p. 40-7.
14. Salaffi, F., et al., The health-related quality of life in rheumatoid arthritis, ankylosing spondylitis, and psoriatic arthritis: a comparison with a selected sample of healthy people. Health Qual Life Outcomes, 2009. 7: p. 25.
15. Lins, L. and F.M. Carvalho, SF-36 total score as a single measure of health-related quality of life: Scoping review. SAGE Open Med, 2016. 4: p. 2050312116671725.
16. John E. Ware, M.K., James E. Dewey, Barbara Gandek, SF36 Health Survey: Manual and Interpretation Guide. 1993.
17. Ware, J.E., Jr., et al., Comparison of methods for the scoring and statistical analysis of SF-36 health profile and summary measures: summary of results from the Medical Outcomes Study. Med Care, 1995. 33(4 Suppl): p. As264-79.
18. McHorney, C.A., J.E. Ware, Jr., and A.E. Raczek, The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care, 1993. 31(3): p. 247-63.
19. Yarlas, A., et al., Psychometric validation of the SF-36(®) Health Survey in ulcerative colitis: results from a systematic literature review. Qual Life Res, 2018. 27(2): p. 273-290.
20. Ding, X., et al., Self-Reported Sleep Apnea and Dementia Risk: Findings from the Prevention of Alzheimer’s Disease with Vitamin E and Selenium Trial. J Am Geriatr Soc, 2016. 64(12): p. 2472-2478.
21. Kryscio, R.J., et al., A randomized controlled Alzheimer’s disease prevention trial’s evolution into an exposure trial: the PREADViSE Trial. J Nutr Health Aging, 2013. 17(1): p. 72-5.
22. Lippman, S.M., et al., Effect of selenium and vitamin E on risk of prostate cancer and other cancers: the Selenium and Vitamin E Cancer Prevention Trial (SELECT). Jama, 2009. 301(1): p. 39-51.
23. Kryscio, R.J., et al., Association of Antioxidant Supplement Use and Dementia in the Prevention of Alzheimer’s Disease by Vitamin E and Selenium Trial (PREADViSE). JAMA Neurol, 2017. 74(5): p. 567-573.
24. Lipton, R.B., et al., Screening for dementia by telephone using the memory impairment screen. J Am Geriatr Soc, 2003. 51(10): p. 1382-90.
25. Buschke, H., et al., Screening for dementia with the memory impairment screen. Neurology, 1999. 52(2): p. 231-8.
26. Morris, J.C., et al., The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part I. Clinical and neuropsychological assessment of Alzheimer’s disease. Neurology, 1989. 39(9): p. 1159-65.
27. de Jager, C.A., M.M. Budge, and R. Clarke, Utility of TICS-M for the assessment of cognitive function in older adults. Int J Geriatr Psychiatry, 2003. 18(4): p. 318-24.
28. Knopman, D.S., et al., Validation of the telephone interview for cognitive status-modified in subjects with normal cognition, mild cognitive impairment, or dementia. Neuroepidemiology, 2010. 34(1): p. 34-42.
29. Galvin, J.E., et al., The AD8: a brief informant interview to detect dementia. Neurology, 2005. 65(4): p. 559-64.
30. Peitsch, L., et al., General life satisfaction predicts dementia in community living older adults: a prospective cohort study. Int Psychogeriatr, 2016. 28(7): p. 1101-9.
31. Gulpers, B., et al., Anxiety as a Predictor for Cognitive Decline and Dementia: A Systematic Review and Meta-Analysis. Am J Geriatr Psychiatry, 2016. 24(10): p. 823-42.
32. Montlahuc, C., et al., Self-rated health and risk of incident dementia: a community-based elderly cohort, the 3C study. Neurology, 2011. 77(15): p. 1457-64.
33. Abner, E.L., et al., Baseline subjective memory complaints associate with increased risk of incident dementia: the PREADVISE trial. J Prev Alzheimers Dis, 2015. 2(1): p. 11-16.
34. Sabia, S., et al., Physical activity, cognitive decline, and risk of dementia: 28 year follow-up of Whitehall II cohort study. BMJ (Clinical research ed.), 2017. 357: p. j2709-j2709.
35. Shimada, H., et al., Cognitive Frailty and Incidence of Dementia in Older Persons. J Prev Alzheimers Dis, 2018. 5(1): p. 42-48.